Uniform Wrappers: Bridging Concave to Quadratizable Functions in Online Optimization

Mohammad Pedramfar, Christopher John Quinn, Vaneet Aggarwal

Keywords: DR-submodular Optimization, Convex Optimization

Abstract:

This paper presents novel contributions to the field of online optimization, particularly focusing on the adaptation of algorithms from concave optimization to more challenging classes of functions. Key contributions include the introduction of uniform wrappers, establishing a vital link between upper-quadratizable functions and algorithmic conversions. Through this framework, the paper demonstrates superior regret guarantees for various classes of up-concave functions under zeroth-order feedback. Furthermore, the paper extends zeroth-order online algorithms to bandit feedback counterparts and offline counterparts, achieving a notable improvement in regret/sample complexity compared to existing approaches.

Primary Area: optimization

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 13051